Best Practices for Agile Data Operations in Internet Finance—Wang Tong

Wang Tong , Vice President of Sales at Beijing Yonghong Business Intelligence Technology Co., Ltd., holds a Master of Engineering from the Beijing University of Aeronautics and Astronautics. He has eight years of experience in product sales and marketing in the field of business intelligence. Previously, he worked for Oracle and IBM, both in important positions in consulting and sales. He has successfully promoted the implementation of multiple large-scale projects and has accumulated rich experience in e-commerce, government, finance, Internet and other industries. Wang Tong is currently mainly responsible for product sales and channel development, and has provided comprehensive data visualization analysis solutions for hundreds of corporate users, including both e-commerce star companies such as Babytree and traditional giants such as China Mobile.

My sharing today mainly includes two parts. One part is about agility, which discusses the best practices of smarter data operations, and will show you cases in the middle. Another important topic is how to build our own data operation system for Internet finance, and what indicators and data analysis methods should be used to guide and help my business decision-making.

Let's talk about the first part first. Now no enterprise dares to say that data has no value to them. Because if you want to build a smart enterprise, data-based operation is the only choice. If an enterprise does not do data-based operation, first of all, it will not have a very intuitive and quantitative perception of the business, and will not have a clear understanding of the business status, development rules, and user portraits. It is equivalent to developing business in the dark by feeling. Second, if there is no data as a basis, we have no way to quickly and effectively solve some emergencies, such as a sudden increase in the amount of funds, an increase in the bad debt rate, etc.; third, without data support, it is difficult to make accurate predictions.

This is a saying from Lord Kelvin, a foreign master: If you can't measure something, you can't improve it. In other words, we judge that this year is better than last year because we have a quantitative standard, not a perceptual indicator.

The Internet finance industry has several technical requirements for data analysis operations.

For the Internet finance industry, the requirements for the underlying information system technology of internal data operations have their own industry characteristics, which we summarize into three types.

The first is the requirement of agility. Because now in the industry, whether it is national or regulatory policies, the market itself, or our business model, there are often rapid changes. This means that our data analysis needs must also be adjusted or changed frequently, which requires agile response. The second is the requirement of high performance. Because the amount of traffic data, transaction data, and user data in the financial field is very large, there is no powerful underlying platform as an effective support for my data analysis, which means that I cannot process the huge amount of data very efficiently, and it may cost a lot. The third is the requirement of self-service. Because all of our friends here are from financial companies, for our financial companies, the characteristic of our personnel ratio may be that more than 90% of the personnel should be financial professionals, or they are all our business personnel, and only 10% of the personnel are IT personnel or technical personnel within our company. Business personnel with analysis needs should be able to personally perform the analysis process to achieve the effect of self-service. Instead of relying entirely on IT personnel for the implementation of all analyses, only in this way can the value of data be truly released, so that everyone in the enterprise can be a data analyst to build a strong enterprise. Therefore, agility, high performance, and self-service are some of the technical requirements that the Internet finance industry has for data analysis and operations.

I just mentioned that the most important thing is the early and mid-term, and the most important thing is the analysis of the entire life cycle. For the entire life cycle, assuming that I am doing P2P, my life cycle should be from the diversion of my advertising channels, to how many people come to my APP or my website, he browses but does not register. The next step is how many people register, but do not recharge. The next step is how many people recharge, but no one bids. And to *** he bids, but he does not re-bid, there is no second or third bid, or only a partial re-bid. I can look at the funnel of my entire life cycle from the number of users, or from the amount, several different angles, and finally form a funnel analysis to look at my traffic, including the conversion of each stage.

For users, it is relatively routine, such as analyzing their attributes and actions, looking at the life cycle value, retention, and bidding preferences of different user segments, matching the preferences of users in each segment, launching my products, and matching transactions.

Having talked so much, we can do these analyses, but when we really start to do it, faced with these massive amounts of data, we definitely need a data analysis system and an IT system to help us meet these analysis needs very flexibly and efficiently.

The technical field of data analysis is also undergoing an international global transformation.

In the past, when we mentioned IT systems for data analysis, we might first think of the well-known giants, such as Oracle's BIEE, IBM's Cognos, SAP's BO, etc. But today, new vendors such as Yonghong Technology, Spotfire, Tableau, and Qlikview are gradually becoming more and more well-known in the industry.

What this represents is a trend. Gartner, one of the most authoritative IT consulting organizations in the world, mentioned a point in its Magic Quadrant report on data analysis this year. The title of this report is that agile and exploratory data analysis has become a general trend. In this report, some traditional data analysis systems such as BIEE and Cognos are classified as the previous generation of data analysis systems. What we are doing now is classified as the new generation of plant spacing analysis systems. The positioning of the previous generation of traditional data analysis systems is that its users are used by IT, so it is a completely IT-centric data analysis system. In recent years, the original previous generation of traditional IT-centric data analysis platforms are increasingly being supplemented or even replaced by this business-driven data analysis platform. In essence, the only difference between the new generation of data analysis platforms and the previous generation of data analysis platforms is the most essential principle, which is to separate the data layer from the logic layer, instead of mixing the data layer and the logic layer together like the traditional data analysis platform.

In summary, through this new generation of agile data analysis platform, we can further unleash the value of data, allowing frontline personnel to flexibly analyze data instead of just letting management look at some fixed reports. It allows an enterprise to go from having a small amount of data to having data that can truly guide everyone's business operations.

At present, domestic companies such as iResearch, Babytree, Tujia.com, and the top-tier P2P JiMuBox are some of our benchmark cases and customers. JiMuBox's risk control, credit investigation, and traffic-end marketing-related analysis are all done through our platform, allowing hundreds of business personnel to perform various real-time analyses every day.